SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017

Complexity Thinking and Cyber-Physical Systems

Boris Brinzer1, Amardeep Banerjee2, Michael Hauth3 #Competence Centre for Modern Production, Mannheim University of Applied Sciences, Germany [email protected]

Abstract — Cyber-physical systems are currently seen as a panacea to global competitiveness in developed Achieving continuous operational excellence in global economies. This work formulates the concept of and local value chains depends upon continuous complexity thinking and discusses practical innovation in products and processes. The emergence implication of complexity in cyber-physical systems. of cyber-physical systems [6], [7], [8] has triggered a Avoidance, prevention and reduction are the existing paradigm change in industrial automation. strategies for managing complexity in production. There is lack of clear consensus in literature and practice with regards to definition, modeling and measurement of complexity in production. In addition Automated systems and the increasing focus on to the prevailing confusion related to the types of automation [9] in industry are seen as one of the complexity, there is an acute scarcity of methods to solutions to be flexible and globally competitive. The empirically measure production complexity. The effects of increasing automation on job simplification following work analyzes the existing complexity for the operator are debatable and yet to be agreed management literature and provides a framework to upon empirically. But certainly, increasing automation manage production complexity. For further use to the at the workplace shifts system complexity towards industry partners, the importance of complexity support personal and system designers. management in extending the lean management framework is discussed. Complexity management is not only dealing with increasing variety in production, as commonly Keywords—complexity in production, lean misunderstood. However, there has been scant focus management, cyber-physical systems, System on organizational [10] and supply chain drivers [11] performance, industry interactions affecting process complexity [2]. The paper aims to motivate complexity based thinking in production and underline the importance of system thinking to deal I. INTRODUCTION with the increasing levels of automation. The constant endeavour to be globally competitive [1] requires efficient and flexible sub-systems that can handle the variety and dynamic interchanging II. OVERVIEW relationships of the system elements. The complexities In the following section, we discuss relevant involved in producing an increasingly high variety of existing concepts and studies forming the basis for products to satisfy dynamic customer demand is a complexity model development. We define cyber- challenge from an economic, social policy and physical systems (CPS) [12] and cyber-physical environmental perspective. production systems (CPPS) [13]. Furthermore, the

importance of determining the appropriate automation Shorter product and technology lifecycles, changing and complexity levels will be discussed. legal regulations and globalized supply chains [1], [2],

[3] have resulted in customer specific production configurations and decentralized networks, adding A. Cyber Physical Systems more complexity [4]. In 2004, it was estimated that in Cyber-physical systems (CPS) are integrations of total 75% of the EU GDP and 70% of employment in computation with physical processes [12]. Embedded Europe were related to manufacturing [5]. computers and networks monitor and control the Although the lines of products and services are physical processes, usually with feedback loops where blurring, the manufacturing industry today is physical processes affect computations and vice witnessing a renaissance especially in Europe [6]. versa.‖ [13].

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 14 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017

systems theory, natural sciences, economics and business management.

CPS as a concept has been investigated in different industry contexts; smart grid [14], virtual production The complexity management approach [20] builds [15] to name a few which underlines the need for upon a number of studies resulting in usage of integrated studies[16] and a concurrent approach from confusing and intertwined terms such as chaotic production, information technology and electronics systems, complex networks and complexity in a industries for effective implementation of CPS. system. Complexity is structural property of a system

Cyber-physical systems in production CPPS [13] are defined as tailor-made (customized) sub-systems which automatize the configuration of the system, in order to have a flexible, agile and efficient production. CPPS are designed to break open the classic automation pyramid and replace it with networked, decentralized or partly self-organizing services.

Usage of ―Transport bugs‖ [9] in warehouse logistics is an example of CPPS and of the resulting increased sophistication in production systems. These ―transport bugs‖ or self-driven goods transporters process information and operate in a decentralized mode.

Furthermore, they might choose their own operation i.e. inherent property which persists in a system even configuration (e.g. forks or containers for in absence of external agents [21]. transportation) and interact with each other to manage Figure 1 shows the common effects of inherent traffic flow. The entire material flow control is spread complexity on the system performance. Fig 1: Motivation for complexity Studies across a number of virtual shoulders. In case of disruption, the bugs react on their own and rectify the problem. Low system performance becomes visible through changes in efficiency, costs and quality, finally This system requires minimum human intervention resulting in panic and radical decisions. Managing but has to work amidst less sophisticated humans who system complexity helps in identifying hidden or already have to deal with an array of challenging and unseen causes of system failure proactively giving sometimes unpredictable problems. The increasing hints on the possible areas of improvement, in order to level of sophistication will drastically change the level avoid post analysis on system degradation. of complexity for shop floor employees, e. g. machine operators and M&R (maintenance and repair) Fig 1: Motivation for complexity Studies personnel. A majority of the authors define and interpret The existing skill sets of employees, existing complexity differently i.e. either as a mathematical management tools and methodologies might be function [22], [23], [24], subjective [25] and perceived inadequate to deal with the increasing system [26] type to name a few. Furthermore the drivers of complexity. complexity are determined using a number of methodologies namely mathematical models [22], B. Understanding Complexity operations research models [23] or an indigenous case specific approach [24], [25], [27]. There is a gap and The studies of the Nobel Prize winner Prigogine lack of consensus in literature for the interpretation, [17], which looked at complexities in biological measurement and management of complexity in systems, invigorated research and renewed the focus production. For a detailed study and summary of the to complexity studies. Complexity management widely used methodologies for measuring complexity, researchers have stated that complexity can either be one can refer to [19], [21] and [28]. controlled or managed but not prevented [18].

The complexity approach which is further discussed

clears the confusion in conceptualizing complexity in Complexity management has origins [19] from

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 15 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017 production and makes it possible for the system one aspect of complexity usually variety reduction manager to link the cause and effect in order to assess [27],[28], [35]. the system behavior as a whole and to investigate the drivers that influence system performance. To summarize, the existing approaches, be it qualitative [25], [27], [28], [35] and quantitative [22], [23],[24] are not evaluated for industry readiness parameters[38], [39] which is important to establish Before understanding the approach to measure trustworthiness, applicability and transferability of a complexity, the link between complexity and levels of research design. automation needs to be investigated. Automation, among other measures, may lead to increase in Although, scholars have studied complexity in productivity [29]. The economic benefit of automation production as early as 1958 [40], but the fact remains, is strong, but is not the only motivation for increasing that system complexity is not exactly detectable, the level of automation. measurable and manageable in its full expression. Therefore a certain degree of openness and creative Satchell [30] defined automation as the replacement thinking is suggested when dealing with complexity in of human activity by machine activities. Parasuraman production. et.al. [31] presented a more complete definition of automation, ―Automation refers to the full or partial replacement of a function previously carried out by the human operator.‖ III. KMP APPROACH TO COMPLEXITY A. Defining Complexity Harlin [32] highlights the problem of realizing an The KMP approach for defining complexity in appropriate level of automation [10]. It is desired that modern production systems takes into account the increases in levels of automation will be effective, but existing theoretical and empirical approaches [19], the actual state of system performance is different [28], [33] and [34]. The KMP model states that the from the expected state due to a number of factors complexity of a system is determined by the following mainly related to complexity. The expected state is the four dimensions i.e. variety, dynamics, interdepen- situation where the full potential of automation is dence and uncertainty (Figure 2). realized and its negative side effects are negligible.

Hence, complexity management will help in determining optimum automation levels which shall result in decreasing human workload, improving process accuracy and worker safety [31].

C. Measuring Complexity in Production In literature, complexity is primarily modeled either through information diversity [22] or an entropy model [35] of the system. Following the analysis by Mattson [25], both the information diversity and entropy model approach, though complementary, are quite abstract, therefore difficult to understand and use in practice.

The mathematical models [22], [35] and [36] although rich in academic rigor, have their limitations. The methodologies used for development of these Fig 2: KMP Definition of Complexity models are not congruent or rather too sophisticated for practical use on the shop floor. Variety includes the number of product and process

variants. Interdependencies focus on the number and Furthermore, empirical evidences and hints on intensity of relations between different elements of a implementing the model for the end user i.e. subsystem, influencing process design and system employees of the shop floor, are missing. structure. Dynamics characterizes the temporal The conceptual models are case specific, for changes in the relationships of the elements. example complexity in software industry [37] or Uncertainty takes into account unknown influences or looking at managing complexity in production through

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 16 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017 unpredictable external drivers, e.g. social or political dimensions and system performance. events. The work is similar to [25] and [26] with respect to the With respect to empirical measurement of analysis for the types of complexity, but differs on the complexity, the challenge lies in designing a holistic identification and the holistic yet universally study which covers all the aspects of complexity and applicable approach for measuring complexity. is easy to implement on the shop floor. The ease of implementation could be measured in terms of time and effort required in measurement of complexity.

Figure 3 gives an overview of the methodology for B. Complexity and Lean Production measuring complexity in production. The KMP complexity approach takes into account the subjective Regardless of the cause, in practice significant com- (intangible), objective (system performance numbers) plexity will exist, and how it is managed will be a key and perceived views of the different stakeholders of to be globally competitive. the system. The total complexity of the system is analyzed using an indigenous approach to understand Furthermore, complexity in production can serve as a the system results and stakeholder mindset. key lever linking manufacturing strategy to organizational policies [10] and work allocation. Complexity management is different from lean management [41], [42] and other performance management [43] tools; it takes into account and encourages the congruence of all the system stakeholders.

The complexity approach urges the designer to be proactive and to identify hidden complexities in the system in order to avoid lack of control and the resulting need for ad-hoc decisions.

The KMP approach suggests that excessive complexity should be considered as a waste hindering value adding activity. As seen in Figure 4, lean production [41] and complexity management both aim towards improving system stability and performance. The advantages and methodology for lean production are known [41]. Fig 3: KMP Methodology to measure complexity

Lean management, i.e. maximum waste reduction with Using the ‗define, measure and manage‘ framework hundred percent value creations, has universal [3], [41] along with our industry partners, we are applications but a clear, concise boundary is missing. currently developing a cockpit of key performance indicators (KPIs) which in part build on existing There are studies [10], [42] which state that the lean performance measures. concept strongly focuses on the managerial perspective; with less faith on the employees viewing The KPIs not only look at standard performance them negatively [44] as mechanical components of a measures, for example OEE [3] (overall equipment production system. effectiveness), but also measure the integration time, the amount of mental load and physical load for new operators on the shop floor. Furthermore, relevant complexity drivers and causes for complexity in the process (rework), tasks (variety) and man (motivation) are being identified. These drivers will be measured, analyzed and correlated with the complexity

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 17 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017

identifies over-complexity and inadequate complexity as a complementary waste of production. Furthermore, it suggests usage of robustness and complexity design concepts to increase overall system performance.

Hence, complexity management in production helps in reducing the negative aspects of lean production and provides a new dimension to the system.

With the advent of CPPS, there would be increasingly complex and sophisticated man-machine interactions. Automatic acquisition, processing and mining of process data would change the quality control and waste reduction mechanisms in production systems.

IV. SUMMARY

The methodology presented in this paper is currently being tested and implemented on shop floors (high volume industrial production and automotive) with the help of objective analysis and subjective interviews. Furthermore, we have developed an assortment of key performances indicators to measure the three types of complexity which can be used to assess the system performance, process robustness and perceived complexity of the production employees.

Fig 4: Same Goal, Different Approaches. Understanding Lean Production and Complexity Management The perceived complexity of the employees is to be measured through subjective interviews investigating the attitude of the operator towards changing workplace, increasing workload.

Lean management, i.e. maximum waste reduction with Using the actionable framework (seen in figure 3 and hundred percent value creations, has universal figure 4) a production manager can define and applications but a clear, concise boundary is missing. measure the complexity in production which can be a

part of a continuous improvement exercise that There are studies [10], [42] which state that the lean simplifies or reduces the existing complexity in a concept strongly focuses on the managerial system. perspective; with less faith on the employees viewing them negatively [44] as mechanical components of a By stating over-complexity is a waste which needs production system. to be managed, we extend the lean management

framework. Our approach helps to look beyond the There is an ongoing debate on the consequences of system results, thereby giving hints on the dimensions lean production. On one hand, lean production is and drivers. This unique approach helps to proactively believed to have negative consequences [15], [45] for understand possible negative effects on the shop floor, the confidence of the employees and their job quality, limit the consequences of over complexity and on the other hand lean production is viewed as the redistribute complexity burden through task means for achieving world-class performance in a allocation, resulting in a robust and manageable humane way with positive effects on employees. system. Hence, depending on the implementer and the work environment, the effect of lean production on work The transferability of academic studies into practice is characteristics and employee outcomes can be both valued through the effective use and implementation positive and negative. of industry academic interactions. Successful imple-

mentation of a complexity management approach Complexity management in production extends the requires active participation and congruence of top lean management framework in multiple ways. First, it management, operators and their supervisors. views the process holistically from an employee, managerial and system perspective. Second, it

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 18 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017

Future research will look at the sources, dimensions and drivers of complexity in different production industries in order to get interesting insights on the commonalities in the control and managing strategies for different manufacturing systems with varying complexity levels.

Further investigations will be made to measure changes in selected KPIs as a consequence of changes in complexity and system parameters in order to test and validate the hypotheses.

With the advent of increasing automation solutions in production [46] and the so called fourth industrial revolution, the existing gap between effect and levels of automation could be marginalized, but this will not be possible without managing the complexity levels of the system and the main driver, i.e. the human being.

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 19 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017

References [20] Levy, David L., ―Applications and Limitations of Complexity Theory in Organization Theory and Strategy,‖ in Handbook [1] Schuh, G., ―Lean Complexity - wachsender Komplexität in of Strategic Management, Second Edition (New York: der Produktion Rechnung tragen,‖ Internal report of Marcel Dekker), pp.67-84, 2000 Laboratory for Machine Tools and Production Engineering, [21] Nabil Abu el Ata, Maurice J. Perks, ―Solving the Dynamic Aachen. [Online]. Available: Complexity Dilemma Predictive and Prescriptive Business http://www.leanmanufacturing.de/de/complexity.html Management: Answering the Need for a New Paradigm‖ (Last accessed on 29.10.16) Springer Berlin Heidelberg, pp. 237-245, 2014 [2] Kersten, W., Koeppen, B., Meyer, C. M. and Kern, E., [22] Sivadasan, S. et.al., ―An information-theoretic methodology ―Reduktion der Prozesskomplexität durch for measuring the operational complexity of Supplier- Modularisierun,‖ Industrie Management 4, pp. 11-14, 2005. customer systems,‖ International Journal of Operations [3] Syska, A.,―Produktionsmanagement: das A - Z wichtiger and Production Management, Vol.22, No.1, pp.80-102, 2002 Methoden und Konzepte für die Produktion von heute,‖. [23] Minakshi Kumari, M S Kulkarni, ―A Complexity Based Look- 1.Auflage. Springer-Verlag, Gabler,Wiesbaden, ahead Mechanism for Shop Floor Decision Making,‖ pp.75-78, 2006 Proceedings of the 48th CIRP Conference on Manufacturing [4] Tücks Gregor et.al. ―Komplexität in der Fertigung Systems, Volume 41, pp.63-68, 2016 beherrschen,‖ Complexity Management Journal 1, Aachen, [24] Changchao Gu et.al, ―Reliability-oriented Complexity pp.9-14, 2011 Analysis of Manufacturing Systems Based on Fuzzy [5] EU High-Level Group report, ―MANUFUTURE 2020 - Axiomatic Domain Mapping,‖ 10th International Conference Assuring the future of manufacturing in Europe,‖ EU on Axiomatic Design (ICAD2016) Volume 53, pp.130–135, Belgium, pp.9-18, November 2004. [Online]. Available: 2016 http://www.manufuture.org (Last accessed on 27.08.16) [25] Mattsson Sandra et.al, ―Validation of the complexity index [6] ACATECH, ―Cyber-Physical System, Driving force for method at three manufacturing companies,‖ IEEE innovation in mobility, health, energy and production,‖ International Symposium on Assembly and Manufacturing Springer-Verlag, Berlin Heidelberg, pp.11-26,December (ISAM), pp.55-57, July 30 -Aug. 2, 2013. 2011 [26] Mattsson, Sandra et. al, ―Perceived production complexity - [7] ACATECH, ―Integrierte Forschungsagenda Cyber-Physical understanding more than parts of a system,‖ International Sytems, ‖ACATECH Studie, pp.297, March 2012 Journal of Production Research, pp. 1–9, 2016 [8] ACATECH, ―Strategic “R&D opportunities for 21st century, [27] Windt K.et.al, ―Complexity cube for characterization of Cyber-physical systems”. Report of the Steering Committee complex production systems,‖ International Journal of for Foundations and Innovation for Cyber-Physical Systems, Computer Integrated Manufacturing, Volume 21 Issue 2, USA, pp.24, January 2013. pp.195-200, March 2008 [9] VDMA, ―Industrie 4.0 konkret - Lösungen für die [28] ElMaraghy, W et.al, ―Complexity in engineering design and industrielle Praxis,‖ Forum Industrie 4.0, pp.3-54, Auflage manufacturing,‖ CIRP Annals - Manufacturing Technology 2016 61 (2), pp. 793–814, 2012 [10] Alvesson, M., and Deetz, S., ―Critical theory and [29] Merchant, M. E., ―The future of Manufacturing,‖ in postmodernism approaches to organizational studies,‖ Handbook of Industrial Automation (Shell, Richard L. and Handbook of Organization Studies 6, Sage, Thousand Oaks, Hall, Ernest L. (Ed)), Marcel Dekker, Inc., Cincinnati. 2000 California, 1996 [30] Satchell, P. ―Innovation and Automation,‖ ed. Ashgate, [11] De Toni A. F at.al., ―Complexity measures in manufacturing USA,1998 systems,” in Proceedings of the European Conference on [31] Parasuraman, R., Sheridan, T. B. and Wickens, C. D., ―A Complex Systems: ”Towards a science of complex systems‖, model for types and levels of human interaction with Internationale Universitaire, Paris (France), pp.1-20, 14-18 automation,‖ IEEE transactions on system, man, and November 2005 cybernetics - Part A: Systems and humans, No. 30, pp. 286- [12] Lee, Edward, ―Cyber Physical Systems: Design Challenges,‖ 296, 2000 Invited Paper, 11th IEEE International Symposium on Object [32] Harlin, U., ―Towards Strategic Disturbance Management in Oriented Real-Time Distributed Computing Advanced Manufacturing Systems,‖ Ph.D. dissertation, (ISORC),Orlando, Florida, USA, 2008 swerea, Chalmers University of Technology, Göteborg, [13] Monostori László, ―Cyber-physical production systems: Sweden, 2000 Roots, expectations and R&D challenges,‖ Variety [33] Melzer, ―Bewertung von Komplexität in der Produktion - Management in Manufacturing. Proceedings of the 47th Analyse und Weiterentwicklung von Ansätzen in der Literatur CIRP Conference on Manufacturing, Systems Procedia CIRP für den Bereich der Produktion,‖ Bachelor Thesis 17, pp.9 – 13, 2014 Colloquium at Mannheim University of Applied Sciences, [14] Khaitan S, McCalley James D., ―Modeling and Simulation of 2015 Network Aspects for Distributed Cyber-Physical Energy [34] A. Banerjee and B. Brinzer ―Complexity Management and Systems,‖ in ―Cyber Physical Systems: A Review. Springer Modern Production Systems‖presentedat the 23rd Int. Conf. Berlin Heidelberg, pp. 1-23, 2015 European Operations Management(EUROMA), Trondheim, [15] Berggren, C., ―Lean production - the end of history, ‖Work Norway, June 17-22, 2016 Employment and Society, Vol. 7 No. 2, pp. 163-188, June [35] Deshmukh A et.al, ―Complexity in manufacturing systems,‖ 1993 Part 1: Analysis of static complexity IIE Transactions, [16] Müller R et.al, ―Planning and developing cyber-physical Volume 30, Issue 7, pp.645-655, September 1999 assembly systems by connecting virtual and real worlds,‖ [36] Calinescu, A. et.al. ―Complexity in Manufacturing: an Changeable, Agile, Reconfigurable & Virtual Production, Information Theoretic Approach,‖ Proceedings of the Procedia CIRP 52, pp. 35 – 40, 2016 International Conference on Complex Systems and [17] Prigogine I., ―From Being to Becoming: time and complexity Complexity in Manufacturing, Warwick, UK, pp.30-44, 2000 in the physical sciences,‖ W.H.Freeman & Co Ltd, San [37] Kluth et. al, ―Evaluation of Complexity Management Systems Francisco, pp.272, 1981 – Systematical and Maturity-based Approach. Variety [18] Lloyd, S. ―Measures of complexity: a non-exhaustive list,‖ Management in Manufacturing‖ Proceedings of the 47th IEEE Control Systems Magazine, 21(4), pp.7–8, 2001 CIRP Conference on Manufacturing Systems, Procedia CIRP [19] Clark, Jason B.; Jacques, David R., ―Practical measurement 17, pp. 224 – 229, 2014 of complexity in dynamic systems,‖ Proceedings of the [38] Bryman, A., ―Social research methods‖ Oxford University Conference on Systems Engineering Research, Volume 8, Press, Oxford, England, 5th Edition, pp. 619-659, 2016 pp.14-21, 2012

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 20 SSRG International Journal of Industrial Engineering (SSRG-IJIE) – volume X Issue Y–Month 2017

[39] Lincoln, Y. S. and Guba, E. G, ―Is Causality a Viable [43] Lohman C et.al., ―Designing a performance measurement Concept?‖ in Naturalistic Inquiry, Sage Publishing, system: A case study,‖ European Journal of Operational Thousand Oaks, USA. 1985. Research 156, pp. 267–286, 2004 [40] Ashby, W. R., ―Requisite variety and its implications for the [44] Kamata, S., ―Japan in the Passing Lane: An Insider’s control of complex systems,‖ Cybernetica, Vol. 1 No. 2, pp. Account of Life in a Japanese Auto Factory,‖ Translated by 83-99, 1958 Tatsuru Akimoto. Pantheon Books, New York, NY, 1983 [41] Ohno Taichi, ―Toyota Production System: Beyond Large- [45] Berggren, C., ―Shop Floor Power and the Dynamics of scale Production,‖ Cambridge, Mass.: Productivity Press, Group Work,‖ in Alternatives to Lean Production, ILR Press, Tokyo, Japan, 1988 Ithaca, United States, pp. 221-257, 1992 [42] Jostein Pettersen, ―Defining lean production: some [46] Frohm, J. K et.al, ―System perspective on task allocation and conceptual and practical issues,‖ The TQM Journal, Vol. 21 automation design,‖ Proceedings of the 8th symposium on Issue 2 pp. 127 – 142, 2009 automated systems based on human skills and knowledge, Göteborg, Sweden, 2003.

ISSN: 2349 - 9362 www.internationaljournalssrg.org Page 21